No spam - just the latest insights!
Join over 30,000 industry professionals who subscribe for free
Mr Dessislav Dobrev
Senior Counsel
Dessislav Dobrev is a leading expert in artificial intelligence and the law and has spearheaded AI matters in both academia and practice. His recent book Artificial Intelligence and the Law: A Comprehensive Guide for the Legal Profession, Academia and Society, published by Thomson Reuters, has been endorsed by high-profile academics, leading law practitioners and global AI leaders. This timely volume is one of the first comprehensive studies of how AI will dramatically affect the law, both as a profession and a regulatory domain, as well as society at large.
As an Adjunct Professor Dessislav created and taught new law school courses on AI at McGill Law School, has previously taught at Georgetown Law School, and has provided seminars on AI at other leading institutions such as the Berkeley Center for Law & Technology and the University of Toronto’s Faculty of Law. In addition, Dessislav serves on the New York State Bar Association’s Task Force on Artificial Intelligence. The task force is dedicated to identifying potential benefits, dangers, and opportunities surrounding AI and to make regulatory recommendations for how to manage and integrate the technology in legal practice.
Mr. Dobrev is educated and has practiced law in both civil and common law jurisdictions and has worked in both public and private sector contexts. He is currently a Senior Counsel at the World Bank Group, where he is exposed to AI issues at the global level. Previously, he practiced corporate law at Davis Polk in New York.
INTERNATIONAL AI IN FINANCIAL SERVICES REVIEW
Recent technological developments have been exponential. At the center of the technological vortex is AI. Socio-economic affairs are entering the realm where machines, through the use of AI, are not merely tools, but entities that can create new tools themselves. As a broad, general-purpose technology, AI promises to have transformative implications for a wide range of areas and its rapid advancement presents diverse challenges. The field of financial services is not impervious and will not be immune to this development. In this context, the delivery of financial services will dramatically change over the foreseeable time horizon. This note will summarize the present state of affairs by focusing on the two key prongs of any AI analytical framework, namely the use cases of AI in the area of financial services and the potential benefits that flow from them, as well as the potential risks associated with them. This will be carried out with a view to current and prospective trends.
Use Cases of AI in Financial Services
Financial services is a data-heavy domain, a trend largely driven by the digitalization of financial services from the development of automated teller machines and online banking to the widespread use of mobile apps. As such, this area is ripe for AI’s exponential growth, as high magnitude of input data, especially structured, is one of the key drivers of AI. Therefore, there has been an increasing scope of use cases and potential applications of AI within the operating models of financial services firms. These use cases are wide-ranging both for external- and internal-facing operations, including credit underwriting, trading and investment, regulatory compliance, risk management, customer service, and back-office operations. AI enables data-driven decisions and the capability of AI to convert large amounts of data into bespoke insights could improve investment decisions, enhance risk assessments and generally aid businesses to spot key trends.
More specifically, the areas of banking, insurance, capital markets and payments are well-equipped to see a growing AI impact. For example, in banking the use of AI in credit risk management is gaining more traction. Decisions regarding the provision of loans can be enhanced by the use of AI to determine the creditworthiness of the borrower by harnessing various data to predict the likelihood of default, thereby improving the accuracy of credit decisions. This, in turn, leads to reducing the rejection of creditworthy customers as well as a decrease of credit losses incurred by financial institutions. As another example, in capital markets and investment management, companies may use AI models to devise investment portfolios and provide clients with real-time insights and trading recommendations. Data analytics with the help of AI can be applied to revenue forecasting, stock price predictions, and even screenings of quarterly earnings calls. A further example of use case of AI is fraud management and detection. This includes AI applications that can seek and identify suspicious conduct or alert of unusual events that could signal fraud. In the insurance domain, there is an increasing use of AI in claims processing and the automation of customer document review. In addition, an overarching use case is AI applications to improve customer support, namely by using AI chatbots or agents, customers could be provided with audio and video responses to their queries. What is more, this can be done through the use of AI agents that can process customer requests and questions and make product recommendations.
Risks of AI in Financial Services
Alongside the use cases and potential benefits, AI in financial services also brings risks. They should be considered, and to the extent possible, mitigated, before deploying AI solutions. There are various types of risks, but some of the key ones to highlight at this time encompass misinformation either intentionally or through embedded AI hallucinations, fraud, privacy violations, explainability, and bias. For example, deepfakes present a clear peril to the integrity of financial transactions and services. This is the ability of generative AI to create and disseminate synthetic content to fraudulently induce participants to undertake targeted actions. Believable imagery or voice can be created with increased ease to mimic and impersonate familiar people. One of the most significant risks is the spread of misinformation, leading to potential market manipulation and fraudulent transactions. This can be a deliberate act, but it could also arise from AI hallucinations that furnish confidently stated but inaccurate output, thereby presenting new challenges for AI management. As another example of risk, the very features that make AI attractive, i.e., speed and efficiency of AI, could, in times of financial crisis, increase systemic financial risk and have destabilizing effects in light of rapid decisions that disregard fast-changing context.
Some mitigation strategies that are being deployed for some of these risks include, for example, retrieval-augmented generation (RAG), a process that enhances the capability of generative AI models by optimizing LLMs. This process uses in-house data repositories (e.g. internal organizational data such as policy or procedural documents) to support the validation and accuracy of responses. Another potential mitigant, not so much a technical one but an organizational one, is the education of companies’ staff on the benefits and risks of AI applications, and overall, by creating new organizational structures, such as embedding a Chief AI Officer as part of the management team.
As a final note, the adoption of AI in financial services is occurring as described above and will continue to expand. However, these processes are still infused with uncertainty. And this is precisely the challenge for stakeholders in financial services. On the one hand, they need to embed AI in their operations, otherwise they may lag behind in efficiency. But at the same time, they need to do so responsibly due to the risks, including reputational ones. Proper governance is needed to match the pace of innovation and the hidden pitfalls of such innovation.
